arfima-figarch
arfima-figarch
hi, everyone, I am looking code for Arfima-figarch model....can anyone help me...
Re: arfima-figarch
That sounds like a gimmick. Have you found any carefully done example of that? (If a paper that uses fractional whatever has no technical discussion of how the fractional calculations are implemented it is not carefully done).
To be more concrete about this, the residual at t from the ARFIMA part of the model is (theoretically) an infinitely long weighted sum of the data, which obviously needs to be truncated somehow. How that is truncated can affect the value quite a bit. Then, the (possibly poorly estimated) residuals are squared and a (theoretically) infinitely long weighted sum of those are needed to compute the FIGARCH variance; that also has to be done as a truncated sum and again, how that is truncated can affect the value quite a bit. If a paper doesn't discuss these issues, then I would strongly recommend that you ignore it (and its results) and search for a technically more complete paper.
To be more concrete about this, the residual at t from the ARFIMA part of the model is (theoretically) an infinitely long weighted sum of the data, which obviously needs to be truncated somehow. How that is truncated can affect the value quite a bit. Then, the (possibly poorly estimated) residuals are squared and a (theoretically) infinitely long weighted sum of those are needed to compute the FIGARCH variance; that also has to be done as a truncated sum and again, how that is truncated can affect the value quite a bit. If a paper doesn't discuss these issues, then I would strongly recommend that you ignore it (and its results) and search for a technically more complete paper.